Abstract
Background
Unpredictable work schedules (UWSs), which are characterized by last-minute schedule changes, on-call shifts, and short advance notice, have become increasingly common and have been linked to disrupted circadian rhythms, impaired sleep homeostasis, and poor health. However, evidence from South Korea, where long hours and low schedule autonomy are prevalent, remains scarce. This study examined the association between UWS and insomnia symptoms among wage workers in South Korea.
Methods
Data from 20,534 wage workers aged 20–64 years were obtained from the 2020 Korean Working Conditions Survey. UWS was defined using three items: sudden callbacks, advance notice of schedule changes, and control over working hours. Insomnia symptoms were measured using the minimal insomnia symptom scale, with scores ≥ 6 indicating insomnia. Weighted multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for insomnia by UWS status after adjusting for demographic, socioeconomic, occupational, and psychosocial factors. Sensitivity analyses applied alternative UWS definitions (sudden recall, combined, strict, and lenient definitions).
Results
The prevalence of insomnia symptoms was 16.10% (157/975) in UWS and 7.25% (1418/19559) in non-UWS (p < 0.001). In the unweighted analysis, UWS was associated with higher odds of insomnia symptoms in the crude model (OR 2.46, 95% CI 2.05–2.94). Moreover, the association remained significant after adjustment for demographic and work-related covariates (Model 1: OR 2.15, 95% CI 1.78–2.59; Model 2: OR 1.64, 95% CI 1.34–2.01). Sensitivity analysis using alternative UWS definitions showed a consistent, statistically significant association in both crude and adjusted models. In the weighted analysis, the association remained significant after full adjustment (OR 2.30, 95% CI 1.74–3.05).
Conclusions
UWS was found to be associated with higher odds of insomnia symptoms among Korean wage workers. Improving the predictability of schedules and ensuring the provision of advance notices may help improve sleep health. Longitudinal studies are needed to clarify causal relationships and to further investigate potential underlying pathways.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12889-026-27761-y.
Keywords: Unpredictable work schedules, Insomnia symptoms, Sleep disturbance, Working conditions, Korean Working Conditions Survey, Wage workers
Background
An unpredictable work schedule (UWS) is a work arrangement in which employees face last-minute schedule changes or receive short advance notice of working hours [1]. UWS has increased with the growth of the service economy, platform-based work, and flexible labor systems [2]. Although predictable scheduling may benefit employers by reducing turnover and improving productivity [1, 3], UWS disrupts family routines, social participation, and rest [1]. Furthermore, UWS can disrupt circadian rhythms and impair restorative sleep [4]. These concerns have attracted substantial international attention to the potential health consequences of UWSs.
Insomnia symptoms are a common health outcome of UWSs. Irregular sleep–wake patterns impair sleep homeostasis and lead to poor sleep cycles [5, 6]. Epidemiological studies show that UWSs are linked to shorter sleep duration and poorer sleep quality [7, 8]. UWSs are also associated with stress, depressive symptoms, anxiety [9, 10], workplace injuries [11], and cardiometabolic disorders [12]. Therefore, sleep problems may represent a potential mechanism underlying the association between UWSs and the aforementioned outcomes.
Although a substantial body of international evidence has documented the detrimental effects of UWS, its health implications have been relatively understudied among workers in South Korea. Most studies have focused on shift work or long working hours [13–15], but the Korean labor market is also characterized by hierarchical organizational structures and low worker autonomy [14], which may limit workers’ ability to adjust their schedules. Studies show that schedule instability is associated with work–family conflict [16], insomnia symptoms in firefighters, and poor sleep and health-related quality of life in adults [17]. Studies from Europe, North America, and Asia also report higher risks of insomnia and sleep–wake disorders in workers subjected to UWSs [3, 8, 18]. These findings highlight the importance of examining how UWS influences sleep health, particularly in the context of the working-time environment in South Korea, where evidence remains scarce.
This study was designed to investigate the association between UWS and insomnia symptoms among wage workers by using the sixth wave of the Korean Working Conditions Survey (KWCS), a national representative survey of the working population. By focusing on schedule unpredictability, which represents a distinct and less-studied dimension of working-time organization, rather than shift work alone, this study provides evidence to inform occupational health policy and research on work schedule instability and sleep health.
Methods
Study population
This study utilized data from the sixth KWCS, which was conducted between October 5, 2020 and April 11, 2021 by the Occupational Safety and Health Research Institute to investigate the working conditions, employment characteristics, and health-related outcomes among the economically active population. Individuals aged ≥ 15 years were recruited from across the country by using a stratified, multistage, random sampling design, and the survey was administered by trained interviewers via structured face-to-face interviews.
A total of 50,538 individuals (23,743 men and 26,795 women) participated in the sixth KWCS. From this initial sample, the sample was restricted to wage workers aged 19–64 years with complete responses for all study variables. Non-wage workers, including self-employed individuals without employees and unpaid family workers (n = 17,475), were excluded. Furthermore, participants with missing or incomplete data were also excluded (n = 10,238), including those lacking information on education, income, or self-rated health (n = 1,462); company size or colleague support (n = 2,843); working duration or weekly working hours (n = 382); shift work, job satisfaction, or job stress (n = 218); schedule change information (n = 55); depression or anxiety (n = 56); and any items related to sleep or UWS (n = 5,222). In addition, 2,291 individuals aged ≥ 65 or < 19 years were excluded because they did not meet the target age criteria. A complete-case analysis was conducted by excluding participants with missing values for any study variable. As there was limited information on excluded participants in the public dataset, detailed comparisons between included and excluded individuals were not feasible. The final analytic sample comprised 20,534 participants (10,197 men and 10,337 women) (Fig. 1).
Fig. 1.
Flowchart of study population selection from the sixth Korean Working Conditions Survey (KWCS)
UWS
UWS was defined according to a previous KWCS-based study [19] using three items reflecting different aspects of schedule unpredictability: (1) frequent sudden callbacks, (2) short advance notice of schedule changes, and (3) low schedule control. Participants were classified as having a UWS if they met at least one of these criteria. The detailed procedure was as follows. First, participants were asked how often they were required to return to work on short notice. Responses of “every day,” “several times per week,” or “several times per month” were classified as UWS. Second, among participants whose working hours regularly changed, advance notice of schedule changes was assessed. Notice on the same day, the day before, or a few days before was classified as UWS. Third, schedule control was evaluated based on the extent to which participants could determine their working hours. Workers whose schedules were entirely determined by their company or restricted to fixed shifts were classified as UWS.
Participants were classified into the non-UWS group only if none of these criteria were met. Participants who reported frequent sudden callbacks were classified into the UWS group regardless of responses to the other items. Nonresponse to the second and third items was treated as indicating stable or self-determined schedules, consistent with prior operational approaches in KWCS-based studies. This multidimensional definition is consistent with prior KWCS-based studies [19] but differs from those used in some international studies that relied on a single indicator.
Insomnia symptoms
Insomnia symptoms were assessed using the minimal insomnia symptom scale (MISS), a validated three-item screening tool for identifying insomnia in population-based studies [20]. Three items from the sixth KWCS, namely, difficulty falling asleep, waking up repeatedly during sleep, and waking up feeling tired and unrefreshed, were used to construct the MISS score. Each item asked participants how often they experienced the symptom during the past 12 months. The response options included “every day,” “several times a week,” “several times a month,” “rarely,” “never,” “don’t know,” and “refuse to answer.” Responses of “don’t know” and “refuse to answer” were treated as missing and excluded from the analysis. Valid responses were scored using a 5-point scale (0 = never, 1 = rarely, 2 = several times a month, 3 = several times a week, 4 = every day). Item scores were summed to generate a total MISS score ranging from 0 to 12, with higher scores indicating more severe insomnia symptoms. On the basis of the original validation study, a cut-off score of six or higher was applied to define insomnia. Participants with MISS scores ≥ 6 and < 6 were categorized as having insomnia symptoms and normal sleep, respectively [20].
Covariates
Potential confounding variables were selected on the basis of prior literature and included sex, age group (20–29, 30–39, 40–49, 50–59, and 60–64), household size (single vs. ≥2 persons), residential area (urban, suburban, and rural), self-rated health (good vs. fair/poor), education (elementary school or lower, middle school, high school, and college or higher), monthly income (KRW < 2, 2–2.99, 3–3.99, and ≥ 4 million), weekly working hours (≤ 18, 18–36, 36–52, and > 52), company size (1–49, 50–299, and ≥ 300 employees), working years (< 5, 5–9, 10–14, and ≥ 15 years), shift work (yes or no), colleague support, job satisfaction, and job stress. Anxiety symptoms were assessed using a single-item question asking whether the respondent had experienced anxiety due to health problems during the past 12 months. Depression was assessed using the World Health Organization-Five Well-Being Index (WHO-5), which consists of five items assessing positive emotional well-being over the past 2 weeks. Each item was scored from 0 (“at no time”) to 5 (“all of the time”), with the total score ranging from 0 to 25 [21, 22]. Participants with WHO-5 scores ≤ 13 were classified as being at risk of depression, whereas those with scores ≥ 14 were classified as having normal emotional well-being.
Statistical analysis
All analyses were conducted using Stata/MP (version 18.0; StataCorp LP, College Station, TX, USA). The complex survey design of the KWCS was accounted for by incorporating stratification, clustering, and sampling weights. Regression analyses were performed using survey-weighted procedures. Descriptive analyses were conducted to compare the general characteristics of the UWS and non-UWS groups by using the chi-square test. Logistic regression analyses were performed to examine the association between UWS and insomnia symptoms (MISS ≥ 6). Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated for three models: (1) crude model; (2) Model 1, which was adjusted for sex, age, education, income, area size, household size, and general health; and (3) Model 2, which was further adjusted for weekly working hours, company size, working-time pattern, job tenure (current job), coworker support, job satisfaction, depression, and anxiety. Both unweighted and weighted samples were analyzed. Sampling weights provided by the KWCS were used for the weighted analyses to obtain population-representative estimates.
A standardized classification of UWS has not yet been established. However, the Sixth European Working Conditions Survey [23] assessed schedule predictability as a subdimension of working-time quality and measured the extent to which workers know their schedules in advance, the frequency of short-notice announcements (same day or the day before), and the incidence of employer requests to work at short notice. The survey revealed that nearly 40% of respondents had been asked to work at short notice in the 12 months preceding the survey. To assess robustness, sensitivity analyses were conducted using the following alternative definitions of UWSs: (1) sudden recall, (2) combined (sudden recall or employer-determined schedules with regular changes and short advance notice [same day or the day before]), (3) strict (sudden recall and employer-determined schedules with regular changes and short advance notice), and (4) lenient (sudden recall or any regular change regardless of the advance-notice period). Additional sensitivity analyses were conducted by excluding shift workers and participants working more than 52 h per week to further assess the robustness of the findings. Statistical significance was set at p < 0.05.
Ethics statement
The data used in this study were obtained from the sixth KWCS, which was conducted between 2020 and 2021 by the Occupational Safety and Health Research Institute. The KWCS is a government-approved national survey registered with Statistics Korea (approval no. 380002) and has been conducted every three years since 2006. All KWCS data are publicly available and fully de-identified to protect the privacy of respondents. Given that this study used secondary data that do not contain any personally identifiable information, the requirement for written informed consent was waived by the Institutional Review Board (IRB) of Catholic Kwandong University International St. Mary’s Hospital (IRB no. IS25IMI0061).
Results
General and work‑related characteristics of the study population
Among the 20,534 participants included in the analysis, 975 (4.75%) and 19,559 (95.25%) were classified into the UWS and non-UWS groups, respectively (Table 1). Frequent sudden callbacks were reported by 3.11% of participants, short advance notice of schedule changes by 1.97%, and low schedule control by 4.37% (Table S1). Compared with non-UWS workers, those in the UWS group were more likely to be men (55.5% vs. 49.4%, p < 0.001), to live alone (23.0% vs. 20.3%, p = 0.043), and to report fair or poor self-rated health (37.6% vs. 24.8%, p < 0.001). There were no significant differences in age distribution or educational attainment, while income distribution showed modest variation (p = 0.036).
Table 1.
General characteristics of the study population by unpredictable work schedule (UWS) classification
| Total (n = 20534) | Non-UWS (n = 19559) | UWS (n = 975) | P-value | |
|---|---|---|---|---|
| Sex (n [%]) | < 0.001 | |||
| Men | 10,197 (49.66) | 9656 (49.37) | 541 (55.49) | |
| Women | 10,337 (50.34) | 9903 (50.63) | 434 (44.51) | |
| Age (years; n [%]) | 0.482 | |||
| 20–29 | 2903 (14.14) | 2746 (14.04) | 157 (16.10) | |
| 30–39 | 5097 (24.82) | 4856 (24.83) | 241 (24.72) | |
| 40–49 | 5565 (27.10) | 5312 (27.16) | 253 (25.95) | |
| 50–59 | 5184 (25.25) | 4944 (25.28) | 240 (24.62) | |
| 60–64 | 1785 (8.69) | 1701 (8.70) | 84 (8.62) | |
| Education (n [%]) | 0.085 | |||
| Elementary school or lower | 220 (1.07) | 204 (1.04) | 16 (1.64) | |
| Middle school | 768 (3.74) | 728 (3.72) | 40 (4.10) | |
| High school | 7061 (34.39) | 6704 (34.28) | 357 (36.62) | |
| College or higher | 12,485 (60.80) | 11,923(60.96) | 562 (57.64) | |
| Income (million Korean Won; n [%]) | 0.036 | |||
| < 2 | 4898 (23.85) | 4659 (23.82) | 239 (24.51) | |
| 2–2.99 | 7510 (36.57) | 7193 (36.78) | 317 (32.51) | |
| 3–3.99 | 4745 (23.11) | 4492 (22.97) | 253 (25.95) | |
| ≥ 4 | 3381 (16.47) | 3215 (16.44) | 166 (17.03) | |
| Region (n [%]) | 0.053 | |||
| Urban | 9532 (46.42) | 9044 (46.24) | 488 (50.05) | |
| Suburban | 8579 (41.77) | 8192 (41.88) | 386 (39.59) | |
| Rural | 2424 (11.80) | 2323 (11.88) | 101 (10.36) | |
| Household (n [%]) | 0.043 | |||
| Single | 4194 (20.42) | 3970 (20.30) | 224 (22.97) | |
| Two or more | 16,340 (79.58) | 15,589 (79.70) | 751 (77.03) | |
| Self-rated health (n [%]) | < 0.001 | |||
| Good | 15,327 (74.64) | 14,719 (75.25) | 608 (62.36) | |
| Fair/poor | 5207 (25.36) | 4840 (24.75) | 367 (37.64) | |
Data are presented as frequency (percentage) as appropriate
Participants in the UWS group were more likely to work over 52 h per week (11.4% vs. 6.3%, p < 0.001), be employed in small companies with fewer than 50 employees (69.0% vs. 64.2%, p = 0.006), and perform shift work (18.5% vs. 9.3%, p < 0.001 [Table 2]). They also more frequently reported low colleague support (10.4% vs. 7.4%, p = 0.001) and low job satisfaction (25.4% vs. 14.5%, p < 0.001). Notably, anxiety symptoms (based on heal_prob5) were substantially more prevalent in the UWS group than in the non-UWS group (13.5% vs. 5.0%, p < 0.001).
Table 2.
Work-related characteristics of the study population by unpredictable work schedule (UWS) classification
| Total (n = 20534) | Non-UWS (n = 19559) | UWS (n = 975) | P-value | |
|---|---|---|---|---|
| Working hours per week (n [%]) | < 0.001 | |||
| ≤ 18 | 746 (3.63) | 693 (3.54) | 53 (5.44) | |
| 18–36 | 2126 (10.35) | 2002 (10.24) | 124 (12.72) | |
| 36–52 | 16,310 (79.43) | 15,623 (79.88) | 687 (70.46) | |
| > 52 | 1352 (6.58) | 1241 (6.34) | 111 (11.38) | |
| Business scale (n [%]) | 0.006 | |||
| 1–49 | 13,227 (64.42) | 12,554 (64.19) | 673 (69.03) | |
| 50–299 | 3740 (18.21) | 3593 (18.37) | 147 (15.08) | |
| ≥ 300 | 3567 (17.37) | 3412 (17.44) | 155 (15.90) | |
| Working years current job (n [%]) | 0.054 | |||
| < 5 | 10,652 (51.87) | 10,154 (51.91) | 498 (51.08) | |
| 5–9 | 4584 (22.32) | 4350 (22.24) | 234 (24.00) | |
| 10–14 | 2512 (12.23) | 2378 (12.16) | 134 (13.74) | |
| ≥ 15 | 2786 (13.57) | 2677 (13.69) | 109 (11.18) | |
| Shift work (n [%]) | < 0.001 | |||
| Yes | 1991 (9.70) | 1811 (9.26) | 180 (18.46) | |
| No | 18,543 (90.30) | 17,748 (90.74) | 795 (81.54) | |
| Colleague support (n [%]) | 0.001 | |||
| Yes | 18,991 (92.49) | 18,117 (92.63) | 874 (89.64) | |
| No | 1543 (7.51) | 1442 (7.37) | 101 (10.36) | |
| Job satisfaction (n [%]) | < 0.001 | |||
| Yes | 17,445 (84.96) | 16,718 (85.47) | 727 (74.56) | |
| No | 3089 (15.04) | 2841 (14.53) | 248 (25.44) | |
| Depression (n [%]) | 0.917 | |||
| Yes | 9823 (47.84) | 9355 (47.83) | 468 (48.00) | |
| No | 10,711 (52.16) | 10,204 (52.17) | 507 (52.00) | |
| Anxiety (n [%]) | < 0.001 | |||
| Yes | 1101 (5.36) | 969 (4.95) | 132 (13.54) | |
| No | 19,433 (94.64) | 18,590 (95.05) | 843 (86.46) | |
Data are presented as mean ± standard deviations or frequency (percentage) as appropriate
Prevalence of insomnia symptoms by UWS status
The prevalence of insomnia (MISS ≥ 6) was significantly higher in the UWS group than in the non-UWS group (16.1% vs. 7.25% in the unweighted sample, respectively; 19.85% vs. 6.68% in the weighted sample, respectively; both p < 0.001 [Table 3]).
Table 3.
Prevalence of insomnia symptoms among study subjects by unpredictable work schedule (UWS) classification
| MISS | Non-UWS (n = 19559) | UWS (n = 975) | P-value | |
|---|---|---|---|---|
|
Unweighted (n = 20534) |
Normal sleep (< 6) | 18,141 (92.75) | 818 (83.90) | < 0.001 |
| Insomnia (≥ 6) | 1418 (7.25) | 157 (16.10) | ||
|
Weighted (n = 24343) |
Normal sleep (< 6) | 22,719 (93.32) | 948 (80.15) | < 0.001 |
| Insomnia (≥ 6) | 1614 (6.68) | 234 (19.85) |
Association between insomnia symptoms and UWS status
In the unweighted sample, UWS was associated with a 2.46-fold higher odds of insomnia in the crude model (95% CI 2.05–2.94 [Table 4]). This association remained significant after further adjusting for demographic and socioeconomic variables in Model 1 (OR 2.15, 95% CI 1.78–2.59) and for occupational, psychosocial, and mental health variables in Model 2 (OR 1.64, 95% CI 1.34–2.01). Female sex, suburban residence, age 50–59 years, job tenure of ≥ 15 years, low job satisfaction, and anxiety symptoms were associated with higher odds of insomnia (Table S2). However, participants with higher education (high school and college or higher), without shift work, and without depression symptoms had lower odds of insomnia.
Table 4.
Association between insomnia symptoms and unpredictable work schedule among study population: unweighted and survey-weighted logistic regression analysis
| Crude | Model 1 | Model 2 | ||||
|---|---|---|---|---|---|---|
| OR (95% CI) | P-value | OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Unweighted | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 2.46 (2.05–2.94) | < 0.001 | 2.15 (1.78–2.59) | < 0.001 | 1.64 (1.34–2.01) | < 0.001 |
| Weighted | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 3.46 (2.65–4.51) | < 0.001 | 3.13 (2.41–4.06) | < 0.001 | 2.30 (1.74–3.05) | < 0.001 |
Model 1: Crude + sex, age, education, income, area, number of households, and health status. Model 2: Model 1 + working hours per week, company scale, work type, work tenure at current job, colleague support, job satisfaction, depression, and anxiety
OR Odds ratio, CI Confidence interval
In the weighted crude model, participants in the UWS group had 3.46 times higher odds of reporting insomnia symptoms (95% CI 2.65–4.51, p < 0.001) than those in the non-UWS group. After adjusting for demographic and socioeconomic factors in Model 1, the association was slightly attenuated but remained significant (OR 3.13, 95% CI 2.41–4.06, p < 0.001). Further adjustment for occupational, psychosocial, and mental health variables in Model 2 also yielded a significant association (OR 2.30, 95% CI 1.74–3.05, p < 0.001). Higher odds of insomnia were associated with female sex, suburban residence, age of 50–59 years, job tenure of ≥ 15 years, low job satisfaction, and anxiety symptoms, whereas lower odds of insomnia were associated with higher educational attainment (high school and college or higher) and good self-rated health (Table S3).
Sensitivity analysis: by UWS definitions
In the unweighted analyses, sudden recall was associated with a higher odds of insomnia symptoms (crude OR 2.65, 95% CI 2.14–3.27 [Table 5]). After adjusting for general and socioeconomic characteristics (Model 1), the association persisted (OR 2.44, 95% CI 1.95–3.05) and remained significant with additional adjustment for occupational, psychosocial, and mental health variables (Model 2: OR 1.79, 95% CI 1.40–2.29). By using the combined definition, an increase in the odds was observed (crude OR 2.62, 95% CI 2.17–3.17; Model 1: OR 2.36, 95% CI 1.93–2.88; Model 2: OR 1.77, 95% CI 1.42–2.20). Under the strict definition, positive associations were also observed (crude OR 4.05, 95% CI 2.38–8.52; Model 1: OR 3.62, 95% CI 1.83–7.16; Model 2: OR 2.31, 95% CI 1.10–4.81). The lenient definition showed a similar pattern (crude OR 2.34, 95% CI 1.97–2.76; Model 1: OR 1.93, 95% CI 1.62–2.29; Model 2: OR 1.44, 95% CI 1.19–1.75). In addition, female sex, age of 30–39 years and 50–59 years, suburban residence, fair/poor self-rated health, large company size (≥ 300 employees), job tenure of ≥ 15 years, and low job satisfaction were associated with higher odds of insomnia symptoms, whereas higher educational attainment (high school and college or higher) and absence of anxiety symptoms were associated with lower odds (Tables S4–S7). These associations were consistent across all UWS definitions and in both unweighted Models 1 and 2.
Table 5.
Association between insomnia symptoms and alternative definitions of unpredictable work schedule among study population: unweighted and survey-weighted logistic regression
| Crude | Model 1 | Model 2 | ||||
|---|---|---|---|---|---|---|
| OR (95% CI) | P-value | OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Unweighted | ||||||
| (A) Sudden recall | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 2.65 (2.14–3.27) | < 0.001 | 2.44 (1.95–3.05) | < 0.001 | 1.79 (1.40–2.29) | < 0.001 |
| (B) Combined | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 2.62 (2.17–3.17) | < 0.001 | 2.36 (1.93–2.88) | < 0.001 | 1.77 (1.42–2.20) | < 0.001 |
| (C) Strict | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 4.05 (2.38–8.52) | < 0.001 | 3.62 (1.83–7.16) | < 0.001 | 2.31 (1.10–4.81) | 0.026 |
| (D) Lenient | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 2.34 (1.97–2.76) | < 0.001 | 1.93 (1.62–2.29) | < 0.001 | 1.44 (1.19–1.75) | < 0.001 |
| Weighted | ||||||
| (A) Sudden recall | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 3.31 (2.49–4.40) | < 0.001 | 3.16 (2.37–4.21) | < 0.001 | 2.26 (1.66–3.08) | < 0.001 |
| (B) Combined | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 3.69 (2.78–4.90) | < 0.001 | 3.45 (2.62–4.54) | < 0.001 | 2.54 (1.88–3.42) | < 0.001 |
| (C) Strict | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 4.97 (2.20–11.22) | < 0.001 | 4.19 (2.03–8.65) | < 0.001 | 2.74 (1.20–6.28) | 0.017 |
| (D) Lenient | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 3.09 (2.40–3.98) | < 0.001 | 2.61 (2.04–3.35) | < 0.001 | 1.92 (1.46–2.53) | < 0.001 |
Model 1: Crude + sex, age, education, income, area, number of households, and health status. Model 2: Model 1 + working hours per week, company scale, work type, work tenure at current job, colleague support, job satisfaction, depression, and anxiety
OR Odds ratio, CI Confidence interval
When design weights were applied, a similar pattern emerged. For sudden recall, the crude OR was 3.31 (95% CI 2.49–4.40) and decreased to 3.16 in Model 1 (95% CI 2.37–4.21) and 2.26 in Model 2 (95% CI 1.66–3.08). Under the combined definition, the associations remained robust (crude OR 3.69, 95% CI 2.78–4.90; Model 1: OR 3.45, 95% CI 2.62–4.54; Model 2: OR 2.54, 95% CI 1.88–3.42). With the strict definition, estimates were larger with wider intervals (crude OR 4.97, 95% CI 2.20–11.22; Model 1: OR 4.19, 95% CI 2.03–8.65; Model 2: OR 2.74, 95% CI 1.20–6.28). The lenient operationalization likewise showed elevated odds (crude OR 3.09, 95% CI 2.40–3.98; Model 1: OR 2.61, 95% CI 2.04–3.35; Model 2 OR 1.92, 95% CI 1.46–2.53; Table 5). In all four definitions of UWS, female, suburban residence, fair/poor self-rated health, large company size (≥ 300 employees), job tenure of ≥ 15 years, and low job satisfaction were associated with higher odds of insomnia symptoms in both Models 1 and 2, whereas higher educational attainment (high school and college or higher) and the absence of anxiety symptoms were associated with lower odds (Tables S8–S11).
Sensitivity analysis: excluding shift workers and those long working hours
Sensitivity analyses excluding shift workers and those working more than 52 h per week yielded results consistent with the main analysis (Table 6). In both analyses, the association between UWS and insomnia symptoms remained significant after excluding shift workers (unweighted OR 1.57, 95% CI 1.25–1.98; weighted OR 2.36, 95% CI 1.71–3.26) and those working more than 52 h per week (unweighted OR 1.64, 95% CI 1.31–2.03; weighted OR 2.34, 95% CI 1.73–3.17) in Model 2. Across these sensitivity analyses, female sex, suburban residence, fair/poor self-rated health, longer job tenure (≥ 15 years), and low job satisfaction were consistently associated with higher odds of insomnia symptoms. Large company size (≥ 300 employees) showed a generally similar tendency across models. In contrast, higher educational attainment and the absence of anxiety symptoms were associated with lower odds (Tables S12–S15).
Table 6.
Association between insomnia symptoms and unpredictable work schedule after excluding shift workers and those working more than 52 h per week: unweighted and survey-weighted logistic regression
| Crude | Model 1 | Model 2 | ||||
|---|---|---|---|---|---|---|
| OR (95% CI) | P-value | OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Unweighted | ||||||
| (A) Excluding shift workers (n = 18543) | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 2.42 (1.98–2.96) | < 0.001 | 2.15 (1.74–2.65) | < 0.001 | 1.57 (1.25–1.98) | < 0.001 |
| (B) Excluding > 52 working hours per week (n = 19182) | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 2.39 (1.97–2.90) | < 0.001 | 2.11 (1.72–2.58) | < 0.001 | 1.64 (1.31–2.03) | < 0.001 |
| Weighted | ||||||
| (A) Excluding shift workers | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 3.56 (2.63–4.83) | < 0.001 | 3.22 (2.39–4.34) | < 0.001 | 2.36 (1.71–3.26) | < 0.001 |
| (B) Excluding > 52 working hours per week | ||||||
| Normal sleep | 1.00 | 1.00 | 1.00 | |||
| Insomnia | 3.38 (2.52–4.53) | < 0.001 | 3.11 (2.33–4.14) | < 0.001 | 2.34 (1.73–3.17) | < 0.001 |
Model 1: Crude + sex, age, education, income, area, number of households, and health status. Model 2: Model 1 + working hours per week, company scale, work type, work tenure at current job, colleague support, job satisfaction, depression, and anxiety
OR Odds ratio, CI Confidence interval
Discussion
This study examined the association between UWS and insomnia symptoms among wage workers using data from the sixth KWCS. Workers exposed to UWS showed higher odds of insomnia symptoms after adjusting for demographic, socioeconomic, occupational, and psychosocial variables. The results were consistent across alternative UWS definitions (sudden recall, combined, strict, and lenient) and when applying a stricter short-notice criterion (same day or the day before). Additional sensitivity analyses excluding shift workers and those working more than 52 h per week yielded consistent results, indicating that the association was not solely explained by these working-time characteristics. Collectively, these findings suggest that schedule unpredictability is associated with insomnia symptoms.
The prevalence of UWS in this study (4.75%) was lower than that reported in previous studies. This difference may reflect variations in operational definitions and study populations, as well as contextual differences in scheduling practices. Nevertheless, the association between UWS and insomnia symptoms was consistently observed across various demographic and occupational subgroups, indicating that its effects are not limited to specific vulnerable populations. Consistent with previous studies, the current study showed a higher prevalence of insomnia symptoms among women [24, 25], individuals with lower educational attainment [26, 27], those reporting fair or poor health [26, 28, 29], and those experiencing anxiety [26]. Considering that insomnia, anxiety, and depression are known to be bidirectionally related [30–32], these conditions may also be closely related, thus underscoring the need for integrated conceptual and therapeutic approaches. Furthermore, low job satisfaction was associated with insomnia symptoms in the current study, in line with prior evidence [33], thus indicating that work-related factors may also contribute importantly to the development of insomnia among workers.
Evidence from other settings also supports the association between UWSs and poor sleep outcomes. Workers with irregular hours show increased odds of moderate and severe sleep disturbance compared with those with regular hours [8], and sleep problems among healthcare professionals have been associated with long working hours, short-notice shifts, and frequent schedule changes [18]. These patterns are consistent with experimental evidence showing that irregular sleep–wake patterns can disrupt circadian rhythms and impair sleep homeostasis and therefore contribute to insomnia symptoms [4–7]. Such findings support the possibility that UWS may be associated with circadian misalignment and insufficient restorative sleep, which may be related to insomnia symptoms. Sleep disturbance may represent a potential pathway linking UWSs to broader adverse outcomes; however, mediation could not be formally assessed due to the cross-sectional study design.
This study extends previous work by documenting the association between UWSs and poor sleep outcomes in a large nationally representative sample of Korean wage workers. Most studies in South Korea have focused on shift work or long working hours [13, 15], and few have assessed schedule unpredictability directly. The Korean labor market is also characterized by hierarchical organizational structures and low worker autonomy [14], which may magnify the effects of schedule unpredictability by limiting opportunities for recovery or negotiation over work hours.
Sleep disruption has been linked to broader adverse outcomes in previous studies, including mental health problems, work–family conflict, and occupational injuries [9–11, 16, 34]. Similar results have been reported in US service-sector workers, where unstable scheduling was linked to poorer self-rated sleep quality, difficulty falling asleep, and more frequent awakenings [1, 2]. Intervention studies in retail settings also found that introducing stable scheduling policies improved workers’ sleep quality and well-being [2, 3], thus supporting the view that schedule predictability could be an actionable target for improving sleep health.
This study has several limitations. First, the cross-sectional design does not allow causal inference; therefore, reverse causality cannot be ruled out. Second, the complete-case analysis approach used in this study may have introduced selection bias, as participants with missing data on UWS or insomnia symptoms were excluded. Third, both exposure and outcome were self-reported, thus possibly introducing reporting bias. However, UWS was defined by following validated approaches from prior studies [16, 18], and insomnia symptoms were assessed using the validated MISS [20]. Fourth, unmeasured factors, such as lifestyle, family responsibilities, or chronotype, may have influenced the findings. Parental schedule instability has been shown to negatively affect children’s sleep [35], highlighting that work-related factors and household context may modify the effects of UWS. Fifth, depression and anxiety may function as confounders, correlates, or potential intermediates depending on the assumed causal structure. Adjusting for these variables may therefore introduce overadjustment if they represent a causal link between UWS and insomnia, although they may also confound the observed association. Future studies should include such contextual factors and use longitudinal or quasi-experimental designs to clarify causal relationships.
Conclusions
UWS was found to be associated with higher odds of insomnia symptoms among wage workers in South Korea. Reducing schedule unpredictability, providing adequate advance notice of shifts, and enhancing worker input in scheduling decisions may be beneficial for sleep health. Further longitudinal evidence is needed to clarify these associations.
Supplementary Information
Supplementary Material 1. Table S1. Prevalence of the non-mutually exclusive components used for unpredictable work schedule (UWS) classification. Table S2. Associations of general and work-related characteristics with insomnia symptoms by unpredictable work schedule classification (unweighted). Table S3. Associations of general and work-related characteristics with insomnia symptoms by unpredictable work schedule classification (weighted). Table S4. Associations of general and work-related characteristics with insomnia symptoms under the sudden recall definition of unpredictable work schedule (unweighted). Table S5. Associations of general and work-related characteristics with insomnia symptoms under the combined definition of unpredictable work schedule (unweighted). Table S6. Associations of general and work-related characteristics with insomnia symptoms under the strict definition of unpredictable work schedule (unweighted). Table S7. Associations of general and work-related characteristics with insomnia symptoms under the lenient definition of unpredictable work schedule (unweighted). Table S8. Associations of general and work-related characteristics with insomnia symptoms under the sudden recall definition of unpredictable work schedule (weighted). Table S9. Associations of general and work-related characteristics with insomnia symptoms under the combined definition of unpredictable work schedule (weighted). Table S10. Associations of general and work-related characteristics with insomnia symptoms under the strict definition of unpredictable work schedule (weighted). Table S11. Associations of general and work-related characteristics with insomnia symptoms under the lenient definition of unpredictable work schedule (weighted). Table S12. Associations of general and work-related characteristics with insomnia symptoms after excluding shift workers (unweighted). Table S13. Associations of general and work-related characteristics with insomnia symptoms after excluding shift workers (weighted). Table S14. Associations of general and work-related characteristics with insomnia symptoms after excluding participants working more than 52 hours per week (unweighted). Table S15. Associations of general and work-related characteristics with insomnia symptoms after excluding participants working more than 52 hours per week (weighted)
Acknowledgements
This research was funded by the Occupational Safety and Health Research Institute.
Abbreviations
- UWS
Unpredictable work schedule
- KWCS
Korean Working Conditions Survey
- MISS
Minimal insomnia symptom scale
- WHO-5
World Health Organization-Five Well-Being Index
- OR
Odds ratios
- CI
95% confidence intervals
- IRB
Institutional Review Board
Authors’ contributions
Conceptualization, YYJ; methodology, YYJ; validation, LJH; formal analysis, YYJ; investigation, LJH and KCD; resources, LJH; data curation, YYJ and LJH; writing—original draft preparation, YYJ, LJH, and KCD; writing—review and editing, YYJ, LJH, and KCD; visualization, YYJ; supervision, YYJ; project administration, LJH; All authors have read and agreed to the published version of the manuscript.
Funding
Not applicable.
Data availability
The datasets used and/or analyzed in the current study are available at https://oshri.kosha.or.kr/oshri/researchField/workingEnvironmentSurvey.do.
Declarations
Ethics approval and consent to participate
The KWCS is a government-approved survey of the National Statistics Office (approval no. 380002). It has been held every 3 years since 2006. The data used in this study were obtained from the sixth KWCS, which was conducted between 2020 and 2021 by the Occupational Safety and Health Research Institute. These data are public, and the personal information of all respondents is de-identified.
The requirement for written informed consent was waived by the IRB of Catholic Kwandong University International St. Mary’s Hospital (IRB no. IS25IMI0061) because this study used secondary data that do not contain any personally identifiable information. All procedures were followed in accordance with the relevant guidelines.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Schneider D, Harknett K. Consequences of routine work schedule instability for worker health and well-being. Am Sociol Rev. 2020;85(2):349–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Harknett K, Schneider D. Unpredictable work schedules and employee well-being. Am Sociol Rev. 2023;88(3):451–78. [Google Scholar]
- 3.Williams JC, Lambert SJ, Kesavan S, Saravanan S, Schneider D. Stable scheduling increases sleep and well-being among service-sector workers. Proc Natl Acad Sci U S A. 2021;118(39):e2107828118.34625472 [Google Scholar]
- 4.Åkerstedt T. Shift work and disturbed sleep/wakefulness. Occup Med (Lond). 2003;53(2):89–94. [DOI] [PubMed] [Google Scholar]
- 5.Dijk DJ, Czeisler CA. Contribution of the circadian pacemaker and the sleep homeostat to sleep propensity, sleep structure, and electroencephalographic slow waves and spindle activity in humans. J Neurosci. 1995;15(5):3526–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Borbély AA, Achermann P. Sleep homeostasis and models of sleep regulation. J Biol Rhythms. 1999;14(6):557–68. [DOI] [PubMed] [Google Scholar]
- 7.Hammig O, Bauer G. Work-life imbalance and mental health among male and female employees in Switzerland. Int J Public Health. 2009;54(2):88–95. [DOI] [PubMed] [Google Scholar]
- 8.Yun S, Lee W, Hwang H, Park H. Irregular work hours and sleep disturbance among Korean service workers. BMC Public Health. 2021;21:1011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kim SY, Lee J, Park Y, Kwon H. Association between schedule unpredictability and depressive symptoms. Occup Environ Med. 2022;79(5):345–52. [Google Scholar]
- 10.Han E, Kim D, Cho H, Park M. Flexible work arrangements and sleep problems in Korean workers. Int J Environ Res Public Health. 2024;21(4):1123.39338006 [Google Scholar]
- 11.Lombardi DA, Folkard S, Willetts JL, Smith GS. Daily sleep, weekly working hours, and risk of work-related injury. Occup Environ Med. 2010;67:538–44. [DOI] [PubMed] [Google Scholar]
- 12.Virtanen M, Ferrie JE, Gimeno D, et al. Long working hours and sleep disturbances: The Whitehall II prospective cohort study. Sleep. 2009;32(6):737–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Lee J, Kim H, Park Y. The association between long working hours and health outcomes in Korea. Korean J Occup Environ Med. 2017;29(4):345–54. [Google Scholar]
- 14.Lee J, Kim HR. Characteristics of working time in South Korea: Long hours, rigid hierarchy, and low autonomy. Korean J Occup Environ Med. 2018;30(1):12–20. [Google Scholar]
- 15.Park H, Choi Y, Kwon J. Shift work and sleep quality among Korean workers. Chronobiol Int. 2020;37(9):1315–24. [Google Scholar]
- 16.Choi SM, Lee Y, Kim J, Park H. Association between unpredictable work schedule and work–family conflict in Korea. Ann Occup Environ Med. 2023;35:e46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Park H, Yun SY, Lee W. Schedule instability and health-related quality of life among Korean adults. BMC Public Health. 2025;25:211.39825311 [Google Scholar]
- 18.Min A, Hong HC. Work schedule characteristics associated with sleep disturbance among healthcare professionals. BMC Nurs. 2022;21:68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Choi SM, Lee Y, Kim J, Park H. Operationalizing unpredictable work schedules in the KWCS: Measurement and classification approach. Ann Occup Environ Med. 2023;35:e46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Broman JE, Smedje H, Mallon L, Hetta J. The Minimal Insomnia Symptom Scale (MISS): A brief measure of sleeping difficulties. Sleep. 2008;31(2):241–6. [DOI] [PubMed] [Google Scholar]
- 21.Bonsignore M, Barkow K, Jessen F, Heun R. Validity of the five-item WHO Well-Being Index (WHO-5) in an elderly population. Eur Arch Psychiatry Clin Neurosci. 2001;251(Suppl 2):II27–31. [DOI] [PubMed] [Google Scholar]
- 22.Krieger T, Zimmermann J, Huffziger S, Ubl B, Diener C, Kuehner C, Grosse Holtforth M. Measuring depression with a well-being index: further evidence for the validity of the WHO Well-Being Index (WHO-5) as a measure of the severity of depression. J Affect Disord. 2014;156:240–4. [DOI] [PubMed] [Google Scholar]
- 23.Eurofound. Sixth European Working Conditions Survey – Overview report (2017 update). Luxembourg: Publications Office of the European Union; 2017.
- 24.Zeng LN, Zong QQ, Yang Y. Gender difference in the prevalence of insomnia: A meta-analysis of observational studies. Front Psychiatry. 2020;11:577429. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Benjafield AV, Kuniyoshi FHS, Malhotra A. Estimation of the global prevalence and burden of insomnia: a systematic literature review-based analysis. Sleep Med Rev. 2025;82:102121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Khaled SM, Petcu C, Al-Thani MA, et al. Prevalence and associated factors of DSM-5 insomnia disorder in the general population in Qatar. BMC Psychiatry. 2021;21(1):84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Chu Z, Xie J, Liu R, et al. Urban-rural differences in the prevalence and influencing factors of insomnia: a cross-sectional study in Foshan, China. Front Psychiatry. 2025;16:1580013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Fu X, Du B, Chen Q, et al. Self-rated health, recovery from work, fatigue, and insomnia among commercial pilots concerning occupational and non-occupational factors. Front Public Health. 2023;10:1050776. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Metwally A, Alalawi AD, Sarrar AAA, et al. Prevalence of insomnia and its associated factors in the general population of Saudi Arabia: A cross sectional study. Cureus. 2023;15(8):e44342. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Jansson-Froejmark M, Lindblom K, et al. A bidirectional relationship between anxiety and depression, and insomnia? A prospective study in the general population. J Psychosom Res. 2008;64(4):443–9. [DOI] [PubMed] [Google Scholar]
- 31.Alvaro PK, Roberts RM, Harris JK. A systematic review assessing bidirectionality between sleep disturbances, anxiety, and depression. Sleep. 2013;36(7):1059–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Peng A, Ji S, Lai W, et al. The bidirectional relationship between sleep disturbance and anxiety: Sleep disturbance is a stronger predictor of anxiety. Sleep Med. 2024;121:63–8. [DOI] [PubMed] [Google Scholar]
- 33.Wels J, Fu R. Associations of employment status, working time and job satisfaction with sleep duration and sleep quality among the 50 + population in Japan. Ind Health. 2025;63(5):465–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Virtanen M, Ferrie JE, Gimeno D, Singh-Manoux A, Elovainio M, Jokela M. Long working hours and sleep disturbances: The Whitehall II prospective cohort study. Sleep. 2009;32(6):737–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Lee S, Kim J, Park E. Parental exposure to work schedule instability and child sleep quality. Sleep Health. 2023;9(6):564–72. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Material 1. Table S1. Prevalence of the non-mutually exclusive components used for unpredictable work schedule (UWS) classification. Table S2. Associations of general and work-related characteristics with insomnia symptoms by unpredictable work schedule classification (unweighted). Table S3. Associations of general and work-related characteristics with insomnia symptoms by unpredictable work schedule classification (weighted). Table S4. Associations of general and work-related characteristics with insomnia symptoms under the sudden recall definition of unpredictable work schedule (unweighted). Table S5. Associations of general and work-related characteristics with insomnia symptoms under the combined definition of unpredictable work schedule (unweighted). Table S6. Associations of general and work-related characteristics with insomnia symptoms under the strict definition of unpredictable work schedule (unweighted). Table S7. Associations of general and work-related characteristics with insomnia symptoms under the lenient definition of unpredictable work schedule (unweighted). Table S8. Associations of general and work-related characteristics with insomnia symptoms under the sudden recall definition of unpredictable work schedule (weighted). Table S9. Associations of general and work-related characteristics with insomnia symptoms under the combined definition of unpredictable work schedule (weighted). Table S10. Associations of general and work-related characteristics with insomnia symptoms under the strict definition of unpredictable work schedule (weighted). Table S11. Associations of general and work-related characteristics with insomnia symptoms under the lenient definition of unpredictable work schedule (weighted). Table S12. Associations of general and work-related characteristics with insomnia symptoms after excluding shift workers (unweighted). Table S13. Associations of general and work-related characteristics with insomnia symptoms after excluding shift workers (weighted). Table S14. Associations of general and work-related characteristics with insomnia symptoms after excluding participants working more than 52 hours per week (unweighted). Table S15. Associations of general and work-related characteristics with insomnia symptoms after excluding participants working more than 52 hours per week (weighted)
Data Availability Statement
The datasets used and/or analyzed in the current study are available at https://oshri.kosha.or.kr/oshri/researchField/workingEnvironmentSurvey.do.